plotEigengeneNetworks WGCNA add ME labels to left side of figure

Question

[normingt]

How do I add ME labels (e.g., MEcyan) to the left hand side of my eigengene adjacency heatmap? I’ve seen this figure produced in publications. Is there an argument for the plotEigengeneNetworks function that I can use?

Here is my script that produces the heatmap with colours for labels, but lacks the text labels:

plotEigengeneNetworks(MEs, "APOE astrocytes eigengene adjacency heatmap", setMargins=TRUE, plotHeatmaps=TRUE, plotDendrograms=FALSE)

I tried colorLabels=FALSE and nothing changed.

I tried ySymbols=names(MEs) which gave an error that the number of ySymbols must match the number of rows in the adjacency matrix.

I tried to figure out how to create the eigengene adjacency matrix on my own so that I could use the WGCNA function labeledHeatmap and have more control, but I could not figure out how to do this.

Thank you!

Louisa

[normingt]

Thanks so much Kevin. I think I’ll try your function. Have a great weekend.

[Kevin Blighe]

Cross-posted: https://support.bioconductor.org/p/126141/

Answer 1

[Kevin Blighe]

I think that the issue is just that particular function, which is a wrapper for multiple other WGCNA functions. You could probably just edit the code for that function and create your own version of it, or perform the module-to-trait correlations by invoking labeledHeatmap() separately, as shown in this part of the WGCNA tutorial: Relating consensus modules to external microarray sample information and exporting network analysis results (opens PDF).

If all else fails, I've got a function on GitHub that can probably work for you, too: https://github.com/kevinblighe/CorLevelPlot

Kevin

[normingt]

Thanks Kevin for your insight and suggestions. Since I do not know C++ to alter the code or have phenotype data, I see that I have 2 options. I can create faux phenotype data, create a labeledHeatmap as detailed the WGCNA tutorial you linked, and use Adobe photoshop to add the ME names to my eigengene adhacency heatmap. However, I am unsure if an edited figure would pass the editorial review process. Therefore my better bet could be to try your correlation heatmap function.

[Kevin Blighe]

Hey, you could try my function, however, for the 'editing' idea, I mean to edit the R code. For example, you can access / view the function code by doing this:

require(WGCNA)
plotEigengeneNetworks
function (multiME, setLabels, letterSubPlots = FALSE, Letters = NULL, 
    excludeGrey = TRUE, greyLabel = "grey", plotDendrograms = TRUE, 
    plotHeatmaps = TRUE, setMargins = TRUE, marDendro = NULL, 
    marHeatmap = NULL, colorLabels = TRUE, signed = TRUE, heatmapColors = NULL, 
    plotAdjacency = TRUE, printAdjacency = FALSE, cex.adjacency = 0.9, 
    coloredBarplot = TRUE, barplotMeans = TRUE, barplotErrors = FALSE, 
    plotPreservation = "standard", zlimPreservation = c(0, 1), 
    printPreservation = FALSE, cex.preservation = 0.9, ...) 
{
    size = checkSets(multiME, checkStructure = TRUE)
    if (!size$structureOK) {
        multiME = fixDataStructure(multiME)
    }
    if (is.null(Letters)) 
        Letters = "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
    if (is.null(heatmapColors)) 
        if (signed) {
            heatmapColors = blueWhiteRed(50)
        }
        else {
            heatmapColors = heat.colors(30)
        }
    nSets = length(multiME)
    cex = par("cex")
    mar = par("mar")
    nPlotCols = nSets
    nPlotRows = as.numeric(plotDendrograms) + nSets * as.numeric(plotHeatmaps)
    if (nPlotRows == 0) 
        stop("Nothing to plot: neither dendrograms not heatmaps requested.")
    par(mfrow = c(nPlotRows, nPlotCols))
    par(cex = cex)
    if (excludeGrey) 
        for (set in 1:nSets) multiME[[set]]$data = multiME[[set]]$data[, 
            substring(names(multiME[[set]]$data), 3) != greyLabel]
    plotPresTypes = c("standard", "hyperbolic", "both")
...

You would then copy that code, edit it, and then create a new function called, e.g., plotEigengeneNetworks_edit() - this can become messy, though. Otherwise, you could just initialise the function parameters yourself and then go through it line by line until you get to the relevant parts.

I'm not sure what you need to change exactly, though.